aiqbench analysis
Counterfactual Behavior Cloning addresses a core bottleneck in offline imitation learning by enabling robots to infer underlying human intent despite noisy or suboptimal demonstrations. By expanding the training dataset with counterfactual actions near observed behaviors, the method effectively filters out unintentional complexity while preserving essential task trends. This advancement allows for more robust skill acquisition from imperfect human teachers without requiring real-time interaction or perfect data collection.
Counterfactual Behavior Cloning: Offline Imitation Learning from Imperfect Human Demonstrations
arXiv:2505.10760v2 Announce Type: replace Abstract: Learning from humans is challenging because people are imperfect teachers. When everyday humans show the robot a new task they want it to perform, humans inevitably make errors (e.g., inputting noisy actions) and provide suboptimal examples (e.g., overshooting the goal). Existing methods often learn by matching some or all of the human's behavior --- but this approach is fundamentally limited because the demonstrations themselves are imperfect. In this work we advance offline imitation learning by enabling robots to extrapolate across nearby actions, instead of only considering what the human actually showed. We achieve this by hypothesizing that all of the human's demonstrations are trying to convey an underlying policy, while the noise and sub-optimality within their behaviors obfuscates the data and introduces unintentional complexity. To recover the underlying policy and learn what the human teacher meant, we introduce Counter-BC, a generalized version of behavior cloning. Counter-BC expands the dataset to include actions close to behaviors the human demonstrated (i.e., counterfactual actions that the human teacher could have intended, but did not actually show). During training Counter-BC autonomously modifies the human's demonstrations within this expanded region to reach a simplified policy that explains the underlying trends in the human's dataset. Theoretically, we prove that Counter-BC can extract a simple and similar-to-demonstration policy from imperfect data, multiple users, and teachers of varying skill levels. Empirically, we compare Counter-BC to state-of-the-art alternatives in settings with noisy demonstrations, standardized datasets, and real human teachers. Overall, we find that trying to extrapolate what the human teacher meant by considering nearby actions can lead to more proficient learning from humans. See videos of our work here: https://youtu.be/XaeOZWhTt68
Abstract:Learning from humans is challenging because people are imperfect teachers. When everyday humans show the robot a new task they want it to perform, humans inevitably make errors (e.g., inputting noisy actions) and provide suboptimal examples (e.g., overshooting the goal). Existing methods often learn by matching some or all of the human's behavior --- but this approach is fundamentally limited because the demonstrations themselves are imperfect. In this work we advance offline imitation learning by enabling robots to extrapolate across nearby actions, instead of only considering what the human actually showed. We achieve this by hypothesizing that all of the human's demonstrations are trying to convey an underlying policy, while the noise and sub-optimality within their behaviors obfuscates the data and introduces unintentional complexity. To recover the underlying policy and learn what the human teacher meant, we introduce Counter-BC, a generalized version of behavior cloning. Counter-BC expands the dataset to include actions close to behaviors the human demonstrated (i.e., counterfactual actions that the human teacher could have intended, but did not actually show). During training Counter-BC autonomously modifies the human's demonstrations within this expanded region to reach a simplified policy that explains the underlying trends in the human's dataset. Theoretically, we prove that Counter-BC can extract a simple and similar-to-demonstration policy from imperfect data, multiple users, and teachers of varying skill levels. Empirically, we compare Counter-BC to state-of-the-art alternatives in settings with noisy demonstrations, standardized datasets, and real human teachers. Overall, we find that trying to extrapolate what the human teacher meant by considering nearby actions can lead to more proficient learning from humans. See videos of our work here: this https URL
Robotics (cs.RO)
arXiv:2505.10760 [cs.RO]
(or arXiv:2505.10760v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2505.10760 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 364 words
Related coverage
SynIL: Leveraging Synergy for Offline Imitation Learning from Imperfect Demonstration Datasets
arXiv · Oct 1, 2026
From Weak Data to Strong Policy: Q-Targets Enable Provable In-Context Reinforcement Learning
arXiv · Sep 28, 2026
DreamFormer: Dream Imitation with a Transformer World Model for Language-Conditioned Robotic Manipulation
arXiv · Oct 6, 2026
When Does On-Policy Interaction Help? Representational Tradeoffs in Value-Based Imitation Learning
arXiv · Aug 3, 2026